基于遗传算法阈值优化的模糊边缘检测

Fuzzy Edge Detection According to Threshold Optimization Based on Genetic Algorithm

针对Pal和King提出的模糊边缘检测算法易导致图像灰度信息丢失的问题,提出一种改进的图像模糊边缘检测算法。算法先使用遗传算法和Otsu得到最佳阈值参数,通过阈值定义一个新的隶属函数将原始图像映射到模糊特征平面;然后利用模糊增强提高区域之间的层次,加强边缘两侧的对比度;再对图像进行灰度增强;最后用Min算子提取出图像的边缘。实验结果表明,改进算法提高了边缘检测质量。

To overcome the drawback of losing of gray information after fuzzy image enhancement in Pal and King algorithm, a modified fuzzy edge detection algorithm is presented in this paper. Firstly, a scheme for threshold using genetic algorithm and Otsu is presented. Secondly, according to the threshold, a new membership function is defined automatically in order to change the original image into the fuzzy property plane. Thirdly, the level of different regions is improved after performing fuzzy enhancement, so that the contrast between both sides of the edge is enhanced. Then, the getting image is enhanced by gray transformation. Finally, the edges of the image are extracted according to Min operator. Experimental results show that the quality of edge detection is improved with the proposed algorithm.